Magnetic Resonance Spectroscopy following Mild Traumatic Brain Injury: A Systematic Review and Meta-Analysis on the Potential to Detect Posttraumatic Neurodegeneration

Traumatic brain injury (TBI) is the most relevant external risk factor for dementia and shares hallmarks of neurodegenerative diseases, such as Alzheimer and Parkinson disease, and frontotemporal dementia [1-3]. Each year, over 50 million people sustain a TBI, which renders TBI the global leading cause of death and disability among the young and working populations. Thus, TBI is a major public health concern with tremendous socioeconomic implications and estimated costs of approximately USD 400 billion annually [4].

Mild TBI (mTBI) accounts for 80–90% of all brain impacts and is defined according to the American Congress of Rehabilitation Medicine (ACRM) by a score of 13–15 on the Glasgow Coma Scale (GCS), a maximum loss of consciousness of 30 min, and posttraumatic amnesia <24 h after the brain injury [5-7]. The term mild TBI often misrepresents the patient’s burden with functional, emotional, and cognitive long-term sequelae, as half of mTBI patients with a negative cranial computed tomography (CCT) complain functional limitations 12 months after the brain impact, and up to one third suffer from cognitive dysfunction [8-12]. Nearly 90% of CCT or routine MRI brain scans remain negative despite major clinical health concerns with the inability to return to work after mTBI [13-17], which emphasizes the need for additional diagnostic tools to identify patients earlier who are at risk for unfavorable outcome or even posttraumatic dementia. Current state-of-the-art CCT and MRI techniques are limited in recognizing the full complexity of the mTBI neuropathology, and thus more sensitive, yet still realistic clinical measures to detect early compromised brain metabolism that might help to predict long-term cognitive deficits are required [16, 18].

Magnetic resonance spectroscopy (MRS) allows for in vivo measurement of metabolites that are undetectable by conventional neuroimaging thereby holding potential to identify mTBI patients that could benefit from specific neuropsychiatric and cognitive rehabilitation [19]. Brain energy metabolism is altered after TBI due to posttraumatic ischemia with mitochondrial dysfunction and loss of neuronal integrity with increased cell membrane turnover. In vivo MRS is an MRI technique that can detect nuclei with spins such as ¹H, an abundant by-product of cellular respiration and brain tissue metabolites. As a noninvasive and safe technique, MRS is available on clinical MR scanners (1.5 and 3.0 T) without ionizing radiation [20]. This method holds the potential to identify compromised brain metabolism, but evidence after mTBI is scarce [21]. Key metabolites such as N-acetyl-aspartate (NAA), glutamate (Glu), and choline (Cho) have the potential to capture the dynamic and fuller picture of the secondary changes occurring in the injured brain, while creatine (Cr) is widely recommended as a reference marker (Table 1). However, its current application in TBI patients is purely analytical [22].

As it stands, there is currently no clinically available objective biological tool for mTBI classification beyond the GCS – albeit a major goal of the large TRACK-TBI and CENTER-TBI trials. Thus, early detection of compromised brain metabolism as marker for cognitive dysfunction or even the risk for posttraumatic dementia would be highly appreciated by clinicians to provide patients and their families with more reliable information, meanwhile creating a standardized protocol for TBI diagnosis [4, 23, 24]. With these conclusions in mind, we performed a systematic review and meta-analysis to elucidate if MRS might be suitable to identify changes in brain metabolism in adult mTBI patients having a negative CCT and/or MRI brain scan during the acute or subacute postinjury phase in comparison to their age- and sex-matched healthy controls (HC).

This systematic review and meta-analysis were performed according to the PRISMA guidelines for systematic reviews and meta-analysis [25, 26]. Eligibility was assessed according to the PICO (patient, intervention, control, outcome) process. We included original data reporting on adult mTBI patients who underwent MRS with an MRI field strength of 1.5 or 3 T within 90 days after injury (acute and subacute phase) and compared those to age- and sex-matched HC. Outcome measures were MRS metabolites measured in different regions of interests (ROIs). mTBI was defined according to the ACRM as an external head and brain impact with a transient or permanent focal neurological deficit and/or any alteration in quantitative or qualitative consciousness that does not exceed (i) a GCS of 13–15 after the first 30 min, (ii) a loss of consciousness of 30 min, and (iii) posttraumatic amnesia <24 h [5, 6].

We included original, English, and peer-reviewed research articles reporting on controlled studies using MRS in adult patients (≥18 years) having experienced a single mTBI (ACMR criteria) during the acute or subacute phase after the brain impact. Data were included from patients with no clinical need for a brain scan or with a negative CCT or MRI. At least 1 of the following metabolites or metabolite ratios had to be measured: NAA, Cho, Glu, Cr, glutamine (Gln), myoinositol (mIns), NAA/Cr, Cho/Cr, Glx (Glu/Gln)/Cr, and mIns/Cr. Exclusion criteria were any brain pathology in a CCT and/or MRI scan, MRS performed beyond 90 days after TBI, children, patients with repetitive concussions with a previous TBI or any premorbidly diagnosed neurobehavioral (e.g., depression or bipolar disorder) or neurological comorbidity (e.g., seizures and brain tumors), non-peer-reviewed articles, conference abstracts, case reports, book chapters, non-controlled studies, interventional studies, animal studies, and non-English language articles.

A comprehensive literature search was conducted from the first edition of electronic databases until January 31, 2020, specifically through PubMed (1901 to present), ScienceDirect (2000 to present), Cochrane Library (1991 to present), and Scopus (1980 to present; Fig. 1). Relevant reviews and their supplementary material were included in full-text screening to check analyzed studies for eligibility as well as their reference lists to ensure that the most possible studies were identified. Search terms are available in the online supplementary Table S1 (for all online suppl. material, see www.karger.com/doi/10.1159/000508098). After removing duplicates, all articles were screened independently in a blinded standardized manner by 2 reviewers (A.E. and M.H.-S.). Disagreements on inclusion between reviewers were resolved for final decision through a detailed discussion and an additional reviewer (K.R.).

A risk of bias assessment was performed using a modified version of the ROBINS-I tool [27]. Evaluation of confounding, selection, classification of intervention(s), missing data, measurement of outcome(s), and selection of the reported result biases were assessed with either low, high, or unclear risk judgments, and the GRADE assessment was used to evaluate the level of evidence across included studies [28].

Demographic data (sex and age) and basic characteristics (MRS time point after TBI, metabolites, ROIs, and MRI field strengths) are presented, and data are given as means ± SD and/or ranges. For the meta-analysis, metabolites were analyzed if applied in at least 2 out of the 4 included studies. Statistical analysis and graphs were performed using Review Manager Software 5.3 (Copenhagen, Denmark) and R (R Core Team, Vienna, Austria), respectively [29]. Effect size (ES) and 95% confidence intervals (CIs) of each defined MRS metabolite (Table 1), namely NAA, Glu, Cho, and Cr, were calculated using a random-effects model according to DerSimonian and Laird [30].

The database search returned 5,395 articles and further 22 articles were identified from relevant reviews or reference lists (Fig. 1). After removing duplicates, 5,276 articles were screened by title and abstract. Thereafter, 94 full-text articles were assessed for eligibility. Two articles were eligible for the systematic review, below indicated as study 1 by Gasparovic et al. [24] and as study 2 by Sivák et al. [31], and 4 articles were eligible for the meta-analysis, namely study 1 [24], study 2 [31], study 3 by Veeramuthu et al. [32], and study 4 by Cohen et al. [33]. Studies 3 and 4 were not eligible for the systematic review [32, 33] due to the lack of subgroup analysis of CCT-negative patients.

The 2 included studies analyzed a total of 31 mTBI patients (23 males, 8 females) and 31 sex- and aged-matched HC (24 males, 7 females) with a mean age of 31 and 30 years, respectively (Table 2). The mean time between TBI and MRS was 6.4 days (range 2.7–19 days). Metabolites quantified in both studies were NAA or NAA + NAA glutamate, NAA/Cr, Cho, and Cr. Authors analyzed the following ROIs: study 1 investigated (i) the white matter (WM) between anterior cingulum, medial frontal gyrus, and superior longitudinal fasciculus, (ii) gray matter (GM) in the interhemispheric fissure, and (iii) the splenium, while study 2 analyzed (iv) the right frontal lobe, (v) the left frontal lobe, and (vi) the pons (Table 2).

Gasparovic et al. [24] (study 1) found a nonsignificant Cho change in different ROIs. In detail, Cho was slightly increased compared to HC in the interhemispheric fissure (GM) and in the splenium, whereas it was decreased in the WM, that is, 1 cm above the lateral ventricles. Cr was significantly increased in the WM and splenium, and nonsignificantly decreased in the GM. Glx was significantly decreased in the GM, and nonsignificantly increased in the WM. No significant effect was seen for NAA in the above-mentioned ROIs and time points, namely 3–19 days after mTBI.

Sivák et al. [31] (study 2) observed a significantly lower NAA concentration in the right and left frontal lobes. Lower NAA/Cr and Cho/Cr ratios were found in the right frontal lobe. After Bonferroni’s adjustment to correct for multiple comparisons, NAA in the left frontal lobe and NAA/Cr in the right frontal lobe remained significantly lower in patients after mTBI compared to HC. Absolute Cho and Cr concentrations in the left frontal lobe tended to be lower in mTBI patients.

Four studies were included in the meta-analysis analyzing 50 mTBI patients (36 males, 14 females) and 51 HC (35 males, 16 females) with a mean age of 31 and 30 years, respectively. Patients received MRS 5.5 days (mean) after the brain impact (Table 3). Of study 3 [32] a subgroup of CCT-negative mTBI patients (n = 12) and of study 4 [33] a subgroup of 7 mTBI and HC were eligible for inclusion. ROIs and their rationales as well as metabolites are detailed in Table 3. Study 2 analyzed the left and right frontal lobes separately, of which we used the data from the left frontal lobe and thereby avoided an overrepresented analysis [31]. Regarding measured metabolites, NAA was analyzed across all 4 studies, while Glu, Cho, and Cr overlapped in 2 out of the 4 studies. MRS sequences, the localization of voxel placement, and voxel sizes differed across all 4 studies or were not given. Studies 1–3 used the PRESS sequence. Study 4 used nonlocalizing proton MRS. Voxel sizes or volumes were 1 × 1 × 1 cm (study 1), 12 mL (cortex), and 9 mL (upper brainstem) (study 2), 19 × 19 × 19 mm (study 3), or were not specified (study 4). Within and across studies, risk of bias was low (online suppl. Table S2).

Across all 4 studies, NAA showed a moderate decline with a moderate overall ES of –0.49 (95% CI –1.08 to 0.09) in 50 mTBI patients compared to 51 HC (Fig. 2). This result was mainly based on the findings of study 2, in which NAA was analyzed in the left frontal lobe in 21 mTBI patients (ES –1.2; 95% CI –1.85 to –0.54). In studies 1, 3, and 4, there were no significant differences in NAA concentrations between mTBI patients and HC.

Studies 1 and 3 found increased Glu, which was primarily measured in the WM of the parietal lobe in mTBI patients compared to HC with an overall strong ES of 0.79 (95% CI 0.17–1.41; Fig. 3).

Studies 1 and 2 reported on decreased Cho in mTBI patients with a weak overall ES of –0.31 (95% CI –0.81 to 0.19; Fig. 4).

Study 1 showed a significant increase in Cr, which was measured in the WM, that is, 1 cm above the lateral ventricles with an ES of 0.84 (95% CI –0.11 to 1.79; Fig. 5). Study 2 reported a nonsignificant Cr decrease in the left frontal lobe with an ES of –0.77 (95% CI –1.4 to –0.15). These contradictory results led to an overall ES for Cr of –0.01 (95% CI –1.59 to 1.58).

In this systematic review and meta-analysis, 2 and 4 out of 5,417 identified records were included, respectively. The most crucial findings were alterations in NAA and Glu concentrations using MRS at 3 or 1.5 T within 90 days after injury, i.e., during the acute and subacute phase after a single mTBI in adults having a CCT/MRI-negative brain scan. Precisely, lowered NAA and increased Glu concentrations were found in mTBI patients compared to HC, indicative for loss of neuronal integrity and disturbed brain metabolism [20, 24, 34]. Cho slightly decreased in mTBI patients, which differs to the amplified Cho concentrations in patients with neurodegenerative diseases and brain tumors as a hallmark of cell membrane turnover [20]. In contrast, there is evidence that Cho increases over time when measured in the WM but not in the GM up to 55 days after mTBI [35], which indicates that the axonal pathology and precise definitions of where and when to measure metabolites are needed. However, the relevance of the overall decreased Cho measured in the GM and WM within study 1 and 2 during the acute and subacute phase after mTBI remains unclear and might be associated to the function of Cho as a precursor molecule for myelin. Thus, we suggest to repetitively measure Cho to elucidate its function over time after mTBI. Results on Cr concentrations in mTBI patients were contradictory and potentially attributed to different field strengths and ROIs. Moreover, a recent longitudinal study demonstrated that higher Cr levels in the subacute stage were positively correlated to cognitive impairment during the chronic stage after mTBI when measured in the centrum semiovale, thus in the WM [36]. Kirov et al. [35] found increasing Cr levels over time when measured in the WM but not in the GM up to 55 days after mTBI, which again emphasizes the axonal pathology after mTBI. In summary, it remains doubtful that Cr is a suitable reference marker following mTBI as previously suggested [20].

When interpreting the current results, different time points of MRS after mTBI must be considered especially as the results of NAA and Cho are mainly driven by study 2 (Sivák et al. [31]), which measured brain metabolites within 33–64 h after mTBI. Thus, early MRS after mTBI could be favorable in detecting relevant metabolic changes, and we suggest bearing this in mind for future investigations.

We chose strict inclusion criteria investigating mTBI patients with negative routine brain scans that are often underdiagnosed yet still exhibit posttraumatic symptoms [37]. Early diagnosis of those presumably mildly brain-injured patients at risk for long-term consequences is critically relevant as longitudinal follow-up studies underline continued symptoms, including chronic fatigue, psychiatric disorders, and unsatisfied quality of life in one third of patients [38-41]. A recent meta-analysis revealed that MRS might have limitations for use in mTBI patients [21], which is contradictory to our findings possibly due to a greater heterogeneity of their study population. In detail, this meta-analysis defined broader inclusion criteria analyzing mTBI not strictly using the ACMR guidelines and additionally included postconcussion patients during the acute, subacute, or chronic phase with or without positive brain scans. This meta-analysis found no brain metabolite changes in these mixed mTBI and postconcussion patients [21], most probably accentuating the different pathophysiology after concussion and single mTBI.

Integrating the current evidence, there is a need for standardized temporal and spatial MRS protocols for future research of mTBI patients.

MRS exploits metabolic alterations in neurodegenerative diseases and brain tumors [42-45], but evidence following mTBI is sparse. Animal studies using MRS after experimental TBI suggest Gln, pyruvate, glycerol, and phosphocholine as suitable metabolites to detect diffuse axonal injury and posttraumatic neurodegeneration [46-50]. However, patients with early signs of dementia have exhibited reductions in Glu and NAA [51], suggesting that these metabolic alterations are not specific for TBI, but might point to the posttraumatic pathway and link to neurodegeneration. Moreover, NAA, Glx, and mIns correlated with cognitive outcomes in pediatric TBI patients of all severities [52]. Repetitive sport-related concussions result in chronic neuroinflammation and neurodegeneration, and MRS was previously suggested as a potential biomarker to detect those neurometabolic changes and cognitive impairment [53-57]. Microglial activation and chronic neuroinflammation are discussed as triggers for posttraumatic neurodegeneration [3, 58-60], and mIns, a metabolite involved in osmoregulation and astrocyte activity, was elevated after mTBI and thus might be a promising marker [61]. Moreover, future studies should focus on the evident metabolic changes in mIns, mIns/Cr, and Glx/Cr in the putamen and, thus, in the deep GM not only involved in motor but also in cognitive and emotional circuits, particularly working memory, cognitive control, category learning, and emotional control, functions often hampered after mTBI [61].

Our systematic review and meta-analysis brought up the heterogeneity of analyzed ROIs following mTBI as the only overlapping ROI was the frontal lobe in 3 out of the 4 studies. In study 2, we included data of the left frontal lobe where NAA was most significantly decreased, and the hampered brain metabolism negatively correlated with neuropsychological performance [31]. Animal studies using MRS following experimental mTBI suggest that the most significant axonal damage is observed in the hippocampus, thalamus, internal capsule, and corpus callosum as early as 2 h after injury [62]. Thus, we suggest these brain regions to be considered in future trials for standardized ROIs after mTBI. To date, it is difficult to draw conclusions from MRS studies because ROIs are not precisely defined and often do not focus on WM well known to be the pathological hallmark of diffuse axonal injury that causes attentional and cognitive deficits after TBI [45].

A limited number of articles with small sample sizes were eligible for inclusion. Furthermore, MRI sequences, field strength, voxel placements, and sizes were incongruent between studies, and Cramer Rao Lower bounds cutoffs were only given in study 3. This is necessary as a good signal depends mostly on voxel size, depth, and on the ROI (e.g., signals are less clear at borders with cerebrospinal fluid or air).

The strongest evidence of metabolic changes was found for decreased NAA and elevated Glu concentrations during the acute to subacute phase in adult mTBI patients compared to HC, while Cr might be not suitable as a reference marker in mTBI patients. If these metabolites have the capacity to recognize patients with negative routine brain scans that are at risk for posttraumatic long-term sequelae is still not yet clear. Follow-up studies correlating MRS with cognitive outcome are necessary to elucidate the potential of these metabolites as biomarkers to predict and detect posttraumatic neurodegeneration at an early stage.

This systematic review and meta-analysis were performed according to the PRISMA guidelines for systematic reviews and meta-analysis. Ethical approval for the literature search was not required.

The authors declare no conflicts of interests.

There are no funding sources to declare.

A.E. and M.H.-S. contributed equally to data collection, analysis, interpretation, and writing of the manuscript. L.M. provided technical expertise on the MRS topic and helped revise the manuscript. K.R. designed and supervised the entire systematic review and meta-analysis, helped in decision making for study inclusion, and wrote and revised the manuscript. All authors approved the final version of the manuscript.

A.E. and M.H.-S. contributed equally to this work.